data mining, ecommerce, fraud detection, and technical
TRANSCRIPT
Data Mining, Data Mining, eCommerceeCommerce, Fraud, FraudDetection, and TechnicalDetection, and Technical
CommunicationCommunicationDonald S. Le Vie, Jr.
Director, Information DevelopmentIntegrated Concepts, Inc.
Austin, Texas
47th Annual Conference of the STC
May 21-24, 2000
Orlando, Florida
OverviewOverview
■ eCommerce, Data Mining, and FraudDetection
■ The Technical Communicator’s Role in all ofThis
■ 3 Examples of eCommerce/Retail Fraud■ Additional Resources
■ Questions
What is What is eCommerceeCommerce??
■ The consumer moves through theInternet to the merchant’s Web site
■ From there, consumer decides topurchase something, so controlmoves to the online transactionserver, where all of the informationthe consumer provides is encrypted
■ Once the order is placed, theinformation moves through a privategateway to a processing network,where the issuing and acquiringbanks complete or deny thetransaction
■ This roundtrip transaction processgenerally occurs in about 5 to 7seconds
OnlineConsumer
MerchantWebSite
World Wide Web Portal
Online Transaction Server
Processing Network
Private Gateway
AcquiringMerchant Bank
IssuingConsumer Bank
What is What is eCommerceeCommerce??
■ eCommerce is not just about Web-basedtransactions
■ Anytime financial transaction data is transmittedelectronically--through ATM machines in the mall,card swipers in the grocery store, or using a creditcard to pay for Disney World tickets--it can be definedas electronic commerce
What is DataWhat is DataMining?Mining?
The process of extracting valid, useful,previously unknown, and ultimatelycomprehensible information from largedatabases and using it to make crucialbusiness decisions, or….”“…torturing the data until it confesses…”
Data Mining MethodsData Mining Methods
■ Association■ Sequence-based
analysis■ Clustering■ Classification
■ Genetic algorithms■ Fuzzy logic■ Estimation■ Fractal-based
transforms■ Neural networks
How Data Mining is Used inHow Data Mining is Used ineCommerceeCommerce
■ Study database schemas■ Identify performance
limitations■ Perform due diligence on
data itself■ Analyze data■ Examine temporal patterns■ Detect, analyze, and mitigate
fraud
■ Examine connection timesrequired for data to passbetween different networknodes
■ Investigate alternativerouting strategies, databasereplication costs, andthroughput
■ Examine metrics associatedwith credit card transactionprocessing at each tasknode within the transactionprocess
The Technical Communicator’s RoleThe Technical Communicator’s Rolein Data Miningin Data Mining
■ Many technical communications professionals are alreadywell-suited for key roles in data mining
■ Solid understanding of databases, statistics and business■ Opportunities in Rapid Application Development (RAD) for
building front-end interfaces that allow users to accesswarehoused data in meaningful ways (Visual Basic,PowerBuilder, client/server, Web-based programming)
■ Conversion and Transformation: experience withstandardizing and conforming information from disparatesystems (can you say SGML? XML?)
■ Project management experience combined with businessand technical savvy to help end users get the most out oftheir data
The Technical Communicator’s RoleThe Technical Communicator’s Rolein in eCommerceeCommerce
■ Establishing eCommerce/Web development proposalmethodologies
■ Writing eCommerce/Web development proposals■ Developing eCommerce project Needs Analysis or
Requirements specifications■ Developing project documentation (Statements of Work,
Functional Design specs, Deliverables specs,Internationalization/Localization specs, userdocumentation plans; user documentation content specs,project post-mortems)
■ Developing online Help■ Understanding issues, practices, and products in the retail
business world and how they translate to eCommerce
Retail Fraud DetectionRetail Fraud Detection
■ Data mining helps businesses account for peak periods ofconsumption, merchandise throughput, irregular transactions
■ Loss attributable to internal fraud is between 40 to 50% of thestock (product) loss suffered by all retail business
■ Data mining can help resolve unusual patterns involvingrefunds, discounts, price overrides, credit cards, store cards,debit cards, staff discounts, voids, reversals, overage andshortages
■ Much of retail fraud occurs through point-of-sale (POS) systemsand doesn’t involve merchandise transactions at all
Types of Retail FraudTypes of Retail Fraud
■ Refund transactions: stolen merchandise presentedwith receipt for a cash refund
■ Discount transactions: “Sweethearting” is offeringspecial discounts to friends and family
■ Skimming: Requires a partner from the outside■ Refunds given to the same account■ Refunds given outside of store hours■ Staff refunds
Refund Given to Same AccountRefund Given to Same Account
1. Blue band (top): accounts2. Green band (bottom): POS operator3. Yellow box: Suspicious activityassociated with this individual4. Dark red line: Predominance ofrefunds being credited to this person’saccount5. Table: base data used to generatedthis diagram (required at least threetransactions per POS operator andaccount number before presentingresults)
The Result: Fired and prosecuted
Refunds Given Before 8 amRefunds Given Before 8 am
Refunds before 8 am made at
these 2 branches
Refunds before8 am made to
these twocredit cards
Operators who maderefunds before8 am
1. Green band: POS Operators2. Red band: Store locations3. Blue band: Credit cards4. Green arrows: Show POSoperators at two store locationsmaking cash refunds prior to 8 am5. Red arrows: Show POS operatorat one store location issuing creditsbefore 8 am to two credit cards
The Problem: Store opens at 10 am
The Result: Fired and prosecuted
Employee Discount FraudEmployee Discount Fraud
OperatorBranch Name
Account
1. Red band: Store branch name2. Green band: Operator name3. Dark blue band: Account numberThe green arrow indicates an operatorat one store branch (red arrow) applied an employee discount to anaccount (blue arrow) in excess of 90% within a 10-minute window (from the table) to an inordinate number of purchases
The result: Fired and probation
Lots of Opportunities to ExploreLots of Opportunities to Explore
■ Data mining and eCommerce offer manyopportunities for technical communicationsprofessionals willing to acquire new skills
■ The difference in value between a subject-matterexpert and a tools expert is increasing
■ The difference in salaries between subject-matterexperts and tools experts is increasing
■ Fraud detection field is ripe with opportunities:incidents of white-collar crime up 30% from 1998.